perf tools: Fix segfault in 'perf top'
PATCH 1/1 perf tools: Fix segfault in 'perf top'
'perf top' segfaults with following operation:
# perf top -e page-faults -p 15173 # 15173 never generate page-fault
Then on the resulting empty interface, press right key:
# ./perf top -e page-faults -p 11400 perf: Segmentation fault -------- backtrace -------- ./perf[0x535428] /lib64/libc.so.6(+0x3545f)[0x7f0dd360745f] ./perf[0x531d46] ./perf(perf_evlist__tui_browse_hists+0x96)[0x5340d6] ./perf[0x44ba2f] /lib64/libpthread.so.0(+0x81d0)[0x7f0dd49dc1d0] /lib64/libc.so.6(clone+0x6c)[0x7f0dd36b90dc]
The bug reside in perf_evsel__hists_browse() that, in the above circumstance browser->selection can be NULL, but code after skip_annotation doesn't consider it.
This patch fix it by checing browser->selection before fetching browser->selection->map and browser->selection->sym.
--- tools/perf/ui/browsers/hists.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)
tools/perf/ui/browsers/hists.c +6 -2
@@ -2017,7 +2017,8 @@ skip_annotation: &options[nr_options], dso); nr_options +=add_map_opt (browser, &actions[nr_options], &options[nr_options], - browser->selection->map); + browser->selection ? + browser->selection->map :NULL ); nr_options +=add_socket_opt (browser, &actions[nr_options], &options[nr_options], socket);
@@ -2030,7 +2031,10 @@ skip_annotation: nr_options +=add_script_opt (browser, &actions[nr_options], &options[nr_options], -NULL , browser->selection->sym); +NULL , + browser->selection ? + browser->selection->sym : +NULL ); } nr_options +=add_script_opt (browser, &actions[nr_options], &options[nr_options],NULL ,NULL );
Signature · 1 line
1.8.3.4
Namhyung Kim
maintainerQuoted 17 lines · Wang Nan
'perf top' segfaults with following operation:
# perf top -e page-faults -p 15173 # 15173 never generate page-faultThen on the resulting empty interface, press right key:
# ./perf top -e page-faults -p 11400 perf: Segmentation fault -------- backtrace -------- ./perf[0x535428] /lib64/libc.so.6(+0x3545f)[0x7f0dd360745f] ./perf[0x531d46] ./perf(perf_evlist__tui_browse_hists+0x96)[0x5340d6] ./perf[0x44ba2f] /lib64/libpthread.so.0(+0x81d0)[0x7f0dd49dc1d0] /lib64/libc.so.6(clone+0x6c)[0x7f0dd36b90dc]
Quoted 6 lines
The bug reside in
perf_evsel__hists_browse()that, in the above circumstancebrowser->selectioncan be NULL, but code afterskip_annotationdoesn't consider it.This patch fix it by checing
browser->selectionbefore fetchingbrowser->selection->mapandbrowser->selection->sym.
I'm ok with this change, just a nitpick below..
Quoted 23 lines
Signed-off-by Wang Nan <[email protected]>Cc Arnaldo Carvalho de Melo <[email protected]>Cc Namhyung Kim <[email protected]>--- tools/perf/ui/browsers/hists.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)tools/perf/ui/browsers/hists.c +2
-1@@ -2017,7 +2017,8 @@ skip_annotation: &options[nr_options], dso); nr_options +=add_map_opt (browser, &actions[nr_options], &options[nr_options], - browser->selection->map); + browser->selection ? + browser->selection->map :NULL ); nr_options +=add_socket_opt (browser, &actions[nr_options], &options[nr_options], socket);
tools/perf/ui/browsers/hists.c skip_annotation: · line 2037
Quoted 8 lines
tools/perf/ui/browsers/hists.c +4
-1@@ -2030,7 +2031,10 @@ skip_annotation: nr_options +=add_script_opt (browser, &actions[nr_options], &options[nr_options], -NULL , browser->selection->sym); +NULL , + browser->selection ? + browser->selection->sym : +NULL );
Is this really needed? IIUC browser->selection is not NULL if browser->he_selection is not NULL. Did you see a case it's NULL? Also passing a NULL thread and a NULL sym is meaningless here since it's a duplication..
Thanks,
Namhyung
Quoted 6 lines
tools/perf/ui/browsers/hists.c +0
-0} nr_options +=add_script_opt (browser, &actions[nr_options], &options[nr_options],NULL ,NULL );Signature · 1 line
1.8.3.4
Quoted 53 lines · Namhyung Kim
Quoted 17 lines · Wang Nan
'perf top' segfaults with following operation:
# perf top -e page-faults -p 15173 # 15173 never generate page-faultThen on the resulting empty interface, press right key:
# ./perf top -e page-faults -p 11400 perf: Segmentation fault -------- backtrace -------- ./perf[0x535428] /lib64/libc.so.6(+0x3545f)[0x7f0dd360745f] ./perf[0x531d46] ./perf(perf_evlist__tui_browse_hists+0x96)[0x5340d6] ./perf[0x44ba2f] /lib64/libpthread.so.0(+0x81d0)[0x7f0dd49dc1d0] /lib64/libc.so.6(clone+0x6c)[0x7f0dd36b90dc]Quoted 6 lines
The bug reside in
perf_evsel__hists_browse()that, in the above circumstancebrowser->selectioncan be NULL, but code afterskip_annotationdoesn't consider it.This patch fix it by checing
browser->selectionbefore fetchingbrowser->selection->mapandbrowser->selection->sym.I'm ok with this change, just a nitpick below..
Quoted 26 lines
Signed-off-by Wang Nan <[email protected]>Cc Arnaldo Carvalho de Melo <[email protected]>Cc Namhyung Kim <[email protected]>--- tools/perf/ui/browsers/hists.c | 8 ++++++-- 1 file changed, 6 insertions(+), 2 deletions(-)tools/perf/ui/browsers/hists.c +2
-2@@ -2017,7 +2017,8 @@ skip_annotation: &options[nr_options], dso); nr_options +=add_map_opt (browser, &actions[nr_options], &options[nr_options], - browser->selection->map); + browser->selection ? + browser->selection->map :NULL ); nr_options +=add_socket_opt (browser, &actions[nr_options], &options[nr_options], socket);@@ -2030,7 +2031,10 @@ skip_annotation: nr_options += add_script_opt(browser, &actions[nr_options], &options[nr_options], -NULL , browser->selection->sym);
Quoted 8 lines
tools/perf/ui/browsers/hists.c skip_annotation: · line 2037
Quoted 4 lines
tools/perf/ui/browsers/hists.c +4
-0+NULL , + browser->selection ? + browser->selection->sym : + NULL);Is this really needed? IIUC
browser->selectionis not NULL ifbrowser->he_selectionis not NULL. Did you see a case it's NULL? Also passing a NULL thread and a NULL sym is meaningless here since it's a duplication..
Actually I didn't see the case when browser->he_selection is not NULL but browser->selection is NULL. I add the testing only for safety reason because we have
if (browser->selection == NULL)
goto skip_annotation;So I guess we'd better to check browser->selection in the codeblock leading by skip_annotation, so if someone breaks the assumption the code still work.
But yes, we can add the checker back when the assumption *does* be broken. Currently let's add a comment here for code readers.
Please have a look for v2 patch.
Thank you.
Quoted 10 lines
Thanks,
NamhyungQuoted 6 lines
tools/perf/ui/browsers/hists.c +0
-0} nr_options +=add_script_opt (browser, &actions[nr_options], &options[nr_options],NULL ,NULL );Signature · 1 line
1.8.3.4
PATCH v2 1/1 perf tools: Fix segfault in 'perf top'
In reply to Namhyung Kim'perf top' segfaults with following operation:
# perf top -e page-faults -p 11400 # 11400 never generates page-fault
Then on the resulting empty interface, press right key:
# ./perf top -e page-faults -p 11400 perf: Segmentation fault -------- backtrace -------- ./perf[0x535428] /lib64/libc.so.6(+0x3545f)[0x7f0dd360745f] ./perf[0x531d46] ./perf(perf_evlist__tui_browse_hists+0x96)[0x5340d6] ./perf[0x44ba2f] /lib64/libpthread.so.0(+0x81d0)[0x7f0dd49dc1d0] /lib64/libc.so.6(clone+0x6c)[0x7f0dd36b90dc]
The bug reside in perf_evsel__hists_browse() that, in the above circumstance browser->selection can be NULL, but code after skip_annotation doesn't consider it.
This patch fix it by checking browser->selection before fetching browser->selection->map.
--- tools/perf/ui/browsers/hists.c | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-)
tools/perf/ui/browsers/hists.c +11 -1
@@ -2017,7 +2017,8 @@ skip_annotation: &options[nr_options], dso); nr_options +=add_map_opt (browser, &actions[nr_options], &options[nr_options], - browser->selection->map); + browser->selection ? + browser->selection->map :NULL ); nr_options +=add_socket_opt (browser, &actions[nr_options], &options[nr_options], socket);
@@ -2027,6 +2028,15 @@ skip_annotation: &actions[nr_options], &options[nr_options], thread, NULL); +/* +* Note that browser->selection != NULL +* when browser->he_selection is not NULL, +* so we don't need to check browser->selection +* before fetching browser->selection->sym like what +* we do before fetching browser->selection->map. +* +* See hist_browser__show_entry. +*/ nr_options += add_script_opt(browser, &actions[nr_options], &options[nr_options],
Signature · 1 line
1.8.3.4
Arnaldo Carvalho de Melo
maintainerEm Mon, Sep 14, 2015 at 10:23:55AM +0000, Wang Nan escreveu:
Quoted 5 lines
'perf top' segfaults with following operation:
# perf top -e page-faults -p 11400 # 11400 never generates page-faultThen on the resulting empty interface, press right key:
So, this happens in perf/urgent, so we need to apply it there first, which this patch doesn't.
I am fixing it up this time, thanks for the patch!
- Arnaldo
Quoted 57 lines
# ./perf top -e page-faults -p 11400 perf: Segmentation fault -------- backtrace -------- ./perf[0x535428] /lib64/libc.so.6(+0x3545f)[0x7f0dd360745f] ./perf[0x531d46] ./perf(perf_evlist__tui_browse_hists+0x96)[0x5340d6] ./perf[0x44ba2f] /lib64/libpthread.so.0(+0x81d0)[0x7f0dd49dc1d0] /lib64/libc.so.6(clone+0x6c)[0x7f0dd36b90dc]The bug reside in
perf_evsel__hists_browse()that, in the above circumstancebrowser->selectioncan be NULL, but code afterskip_annotationdoesn't consider it.This patch fix it by checking
browser->selectionbefore fetchingbrowser->selection->map.Signed-off-by Wang Nan <[email protected]>Cc Arnaldo Carvalho de Melo <[email protected]>Cc Namhyung Kim <[email protected]>--- tools/perf/ui/browsers/hists.c | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-)tools/perf/ui/browsers/hists.c +11
-1@@ -2017,7 +2017,8 @@ skip_annotation: &options[nr_options], dso); nr_options +=add_map_opt (browser, &actions[nr_options], &options[nr_options], - browser->selection->map); + browser->selection ? + browser->selection->map :NULL ); nr_options +=add_socket_opt (browser, &actions[nr_options], &options[nr_options], socket);@@ -2027,6 +2028,15 @@ skip_annotation: &actions[nr_options], &options[nr_options], thread, NULL); +/* +* Note that browser->selection != NULL +* when browser->he_selection is not NULL, +* so we don't need to check browser->selection +* before fetching browser->selection->sym like what +* we do before fetching browser->selection->map. +* +* See hist_browser__show_entry. +*/ nr_options += add_script_opt(browser, &actions[nr_options], &options[nr_options],Signature · 1 line
1.8.3.4
Arnaldo Carvalho de Melo
maintainerEm Mon, Sep 14, 2015 at 03:07:14PM -0300, Arnaldo Carvalho de Melo escreveu:
Quoted 11 lines
Em Mon, Sep 14, 2015 at 10:23:55AM +0000, Wang Nan escreveu:
Quoted 5 lines
'perf top' segfaults with following operation:
# perf top -e page-faults -p 11400 # 11400 never generates page-faultThen on the resulting empty interface, press right key:
So, this happens in perf/urgent, so we need to apply it there first, which this patch doesn't.
I am fixing it up this time, thanks for the patch!
For reference, below is the resulting patch, pushed to acme/perf/urgent, now checking if there are more patches to go together with this one...
- Arnaldo
tools/perf/ui/browsers/hists.c +11 -1
@@ -1968,7 +1968,8 @@ skip_annotation: &options[nr_options], dso); nr_options +=add_map_opt (browser, &actions[nr_options], &options[nr_options], - browser->selection->map); + browser->selection ? + browser->selection->map :NULL );/* perf script support */ if (browser->he_selection) {
@@ -1976,6 +1977,15 @@ skip_annotation: &actions[nr_options], &options[nr_options], thread, NULL); +/* +* Note that browser->selection != NULL +* when browser->he_selection is not NULL, +* so we don't need to check browser->selection +* before fetching browser->selection->sym like what +* we do before fetching browser->selection->map. +* +* See hist_browser__show_entry. +*/ nr_options += add_script_opt(browser, &actions[nr_options], &options[nr_options],
tip-bot for Wang Nan
merge bot[tip:perf/core] perf top: Fix segfault pressing -> with no hist entries
In reply to Wang Nanperf top: Fix segfault pressing -> with no hist entries
Commit-ID: bd315aab8a3ab1bc7074774b89a5d8ec7c1ff7ab Gitweb: http://git.kernel.org/tip/bd315aab8a3ab1bc7074774b89a5d8ec7c1ff7ab Author: Wang Nan <[email protected]> AuthorDate: Mon, 14 Sep 2015 10:23:55 +0000 Committer: Arnaldo Carvalho de Melo <[email protected]> CommitDate: Mon, 14 Sep 2015 15:10:41 -0300
perf top: Fix segfault pressing -> with no hist entries
'perf top' segfaults with following operation:
# perf top -e page-faults -p 11400 # 11400 never generates page-fault
Then on the resulting empty interface, press right key:
# ./perf top -e page-faults -p 11400 perf: Segmentation fault -------- backtrace -------- ./perf[0x535428] /lib64/libc.so.6(+0x3545f)[0x7f0dd360745f] ./perf[0x531d46] ./perf(perf_evlist__tui_browse_hists+0x96)[0x5340d6] ./perf[0x44ba2f] /lib64/libpthread.so.0(+0x81d0)[0x7f0dd49dc1d0] /lib64/libc.so.6(clone+0x6c)[0x7f0dd36b90dc]
The bug resides in perf_evsel__hists_browse() that, in the above circumstance browser->selection can be NULL, but code after skip_annotation doesn't consider it.
This patch fix it by checking browser->selection before fetching browser->selection->map.
--- tools/perf/ui/browsers/hists.c | 12 +++++++++++- 1 file changed, 11 insertions(+), 1 deletion(-)
tools/perf/ui/browsers/hists.c +11 -1
@@ -1968,7 +1968,8 @@ skip_annotation: &options[nr_options], dso); nr_options +=add_map_opt (browser, &actions[nr_options], &options[nr_options], - browser->selection->map); + browser->selection ? + browser->selection->map :NULL );/* perf script support */ if (browser->he_selection) {
@@ -1976,6 +1977,15 @@ skip_annotation: &actions[nr_options], &options[nr_options], thread, NULL); +/* +* Note that browser->selection != NULL +* when browser->he_selection is not NULL, +* so we don't need to check browser->selection +* before fetching browser->selection->sym like what +* we do before fetching browser->selection->map. +* +* See hist_browser__show_entry. +*/ nr_options += add_script_opt(browser, &actions[nr_options], &options[nr_options],